Pull uaccess/access_ok updates from Al Viro:
"Removals of trivially pointless access_ok() calls.
Note: the fiemap stuff was removed from the series, since they are
duplicates with part of ext4 series carried in Ted's tree"
* 'uaccess.access_ok' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs:
vmci_host: get rid of pointless access_ok()
hfi1: get rid of pointless access_ok()
usb: get rid of pointless access_ok() calls
lpfc_debugfs: get rid of pointless access_ok()
efi_test: get rid of pointless access_ok()
drm_read(): get rid of pointless access_ok()
via-pmu: don't bother with access_ok()
drivers/crypto/ccp/sev-dev.c: get rid of pointless access_ok()
omapfb: get rid of pointless access_ok() calls
amifb: get rid of pointless access_ok() calls
drivers/fpga/dfl-afu-dma-region.c: get rid of pointless access_ok()
drivers/fpga/dfl-fme-pr.c: get rid of pointless access_ok()
cm4000_cs.c cmm_ioctl(): get rid of pointless access_ok()
nvram: drop useless access_ok()
n_hdlc_tty_read(): remove pointless access_ok()
tomoyo_write_control(): get rid of pointless access_ok()
btrfs_ioctl_send(): don't bother with access_ok()
fat_dir_ioctl(): hadn't needed that access_ok() for more than a decade...
dlmfs_file_write(): get rid of pointless access_ok()
xdp_umem.c had overlapping changes between the 64-bit math fix
for the calculation of npgs and the removal of the zerocopy
memory type which got rid of the chunk_size_nohdr member.
The mlx5 Kconfig conflict is a case where we just take the
net-next copy of the Kconfig entry dependency as it takes on
the ESWITCH dependency by one level of indirection which is
what the 'net' conflicting change is trying to ensure.
Signed-off-by: David S. Miller <davem@davemloft.net>
net/mlx5: Add ability to read and write ECE options
net/mlx5: Add support for RDMA TX FT headers modifying
net/mlx5: Move iseg access helper routines close to mlx5_core driver
net/mlx5: Cleanup mlx5_ifc_fte_match_set_misc2_bits
net/mlx5: Add support in forward to namespace
{IB/net}/mlx5: Simplify don't trap code
net/mlx5: Replace zero-length array with flexible-array
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Allow a ULP to ask the core to provide a completion queue based on a
least-used search on a per-device CQ pools. The device CQ pools grow in a
lazy fashion when more CQs are requested.
This feature reduces the amount of interrupts when using many QPs. Using
shared CQs allows for more effcient completion handling. It also reduces
the amount of overhead needed for CQ contexts.
Test setup:
Intel(R) Xeon(R) Platinum 8176M CPU @ 2.10GHz servers.
Running NVMeoF 4KB read IOs over ConnectX-5EX across Spectrum switch.
TX-depth = 32. The patch was applied in the nvme driver on both the target
and initiator. Four controllers are accessed from each core. In the
current test case we have exposed sixteen NVMe namespaces using four
different subsystems (four namespaces per subsystem) from one NVM port.
Each controller allocated X queues (RDMA QPs) and attached to Y CQs.
Before this series we had X == Y, i.e for four controllers we've created
total of 4X QPs and 4X CQs. In the shared case, we've created 4X QPs and
only X CQs which means that we have four controllers that share a
completion queue per core. Until fourteen cores there is no significant
change in performance and the number of interrupts per second is less than
a million in the current case.
==================================================
|Cores|Current KIOPs |Shared KIOPs |improvement|
|-----|---------------|--------------|-----------|
|14 |2332 |2723 |16.7% |
|-----|---------------|--------------|-----------|
|20 |2086 |2712 |30% |
|-----|---------------|--------------|-----------|
|28 |1971 |2669 |35.4% |
|=================================================
|Cores|Current avg lat|Shared avg lat|improvement|
|-----|---------------|--------------|-----------|
|14 |767us |657us |14.3% |
|-----|---------------|--------------|-----------|
|20 |1225us |943us |23% |
|-----|---------------|--------------|-----------|
|28 |1816us |1341us |26.1% |
========================================================
|Cores|Current interrupts|Shared interrupts|improvement|
|-----|------------------|-----------------|-----------|
|14 |1.6M/sec |0.4M/sec |72% |
|-----|------------------|-----------------|-----------|
|20 |2.8M/sec |0.6M/sec |72.4% |
|-----|------------------|-----------------|-----------|
|28 |2.9M/sec |0.8M/sec |63.4% |
====================================================================
|Cores|Current 99.99th PCTL lat|Shared 99.99th PCTL lat|improvement|
|-----|------------------------|-----------------------|-----------|
|14 |67ms |6ms |90.9% |
|-----|------------------------|-----------------------|-----------|
|20 |5ms |6ms |-10% |
|-----|------------------------|-----------------------|-----------|
|28 |8.7ms |6ms |25.9% |
|===================================================================
Performance improvement with sixteen disks (sixteen CQs per core) is
comparable.
Link: https://lore.kernel.org/r/1590568495-101621-3-git-send-email-yaminf@mellanox.com
Signed-off-by: Yamin Friedman <yaminf@mellanox.com>
Reviewed-by: Or Gerlitz <ogerlitz@mellanox.com>
Reviewed-by: Max Gurtovoy <maxg@mellanox.com>
Reviewed-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
kobject_init_and_add() takes reference even when it fails. If this
function returns an error, kobject_put() must be called to properly clean
up the memory associated with the object. Previous
commit b8eb718348 ("net-sysfs: Fix reference count leak in
rx|netdev_queue_add_kobject") fixed a similar problem.
Link: https://lore.kernel.org/r/20200528030231.9082-1-wu000273@umn.edu
Signed-off-by: Qiushi Wu <wu000273@umn.edu>
Reviewed-by: Jason Gunthorpe <jgg@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
The mlx5 VF driver doesn't set QP tx port affinity because it doesn't know
if the lag is active or not, since the "lag_active" works only for PF
interfaces. In this case for VF interfaces only one lag is used which
brings performance issue.
Add a lag_tx_port_affinity CAP bit; When it is enabled and
"num_lag_ports > 1", then driver always set QP tx affinity, regardless
of lag state.
Link: https://lore.kernel.org/r/20200527055014.355093-1-leon@kernel.org
Signed-off-by: Mark Zhang <markz@mellanox.com>
Reviewed-by: Maor Gottlieb <maorg@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Since the session name by itself is not sufficient to uniquely identify a
queue pair, include the queue pair number. Show the ASCII channel state
name instead of the numeric value. This change makes the ib_srpt debug
output more consistent.
Link: https://lore.kernel.org/r/20200525172212.14413-3-bvanassche@acm.org
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
pin_user_pages_fast() doesn't need that from its caller.
NB: only reachable from ->ioctl(), and only under USER_DS
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Add a helper to directly set the TCP_NODELAY sockopt from kernel space
without going through a fake uaccess. Cleanup the callers to avoid
pointless wrappers now that this is a simple function call.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Acked-by: Sagi Grimberg <sagi@grimberg.me>
Acked-by: Jason Gunthorpe <jgg@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add a helper to directly set the SO_REUSEADDR sockopt from kernel space
without going through a fake uaccess.
For this the iscsi target now has to formally depend on inet to avoid
a mostly theoretical compile failure. For actual operation it already
did depend on having ipv4 or ipv6 support.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Acked-by: Sagi Grimberg <sagi@grimberg.me>
Signed-off-by: David S. Miller <davem@davemloft.net>
When connected mode is set, and we have connected and datagram traffic in
parallel, ipoib might crash with double free of datagram skb.
The current mechanism assumes that the order in the completion queue is
the same as the order of sent packets for all QPs. Order is kept only for
specific QP, in case of mixed UD and CM traffic we have few QPs (one UD and
few CM's) in parallel.
The problem:
----------------------------------------------------------
Transmit queue:
-----------------
UD skb pointer kept in queue itself, CM skb kept in spearate queue and
uses transmit queue as a placeholder to count the number of total
transmitted packets.
0 1 2 3 4 5 6 7 8 9 10 11 12 13 .........127
------------------------------------------------------------
NL ud1 UD2 CM1 ud3 cm2 cm3 ud4 cm4 ud5 NL NL NL ...........
------------------------------------------------------------
^ ^
tail head
Completion queue (problematic scenario) - the order not the same as in
the transmit queue:
1 2 3 4 5 6 7 8 9
------------------------------------
ud1 CM1 UD2 ud3 cm2 cm3 ud4 cm4 ud5
------------------------------------
1. CM1 'wc' processing
- skb freed in cm separate ring.
- tx_tail of transmit queue increased although UD2 is not freed.
Now driver assumes UD2 index is already freed and it could be used for
new transmitted skb.
0 1 2 3 4 5 6 7 8 9 10 11 12 13 .........127
------------------------------------------------------------
NL NL UD2 CM1 ud3 cm2 cm3 ud4 cm4 ud5 NL NL NL ...........
------------------------------------------------------------
^ ^ ^
(Bad)tail head
(Bad - Could be used for new SKB)
In this case (due to heavy load) UD2 skb pointer could be replaced by new
transmitted packet UD_NEW, as the driver assumes its free. At this point
we will have to process two 'wc' with same index but we have only one
pointer to free.
During second attempt to free the same skb we will have NULL pointer
exception.
2. UD2 'wc' processing
- skb freed according the index we got from 'wc', but it was already
overwritten by mistake. So actually the skb that was released is the
skb of the new transmitted packet and not the original one.
3. UD_NEW 'wc' processing
- attempt to free already freed skb. NUll pointer exception.
The fix:
-----------------------------------------------------------------------
The fix is to stop using the UD ring as a placeholder for CM packets, the
cyclic ring variables tx_head and tx_tail will manage the UD tx_ring, a
new cyclic variables global_tx_head and global_tx_tail are introduced for
managing and counting the overall outstanding sent packets, then the send
queue will be stopped and waken based on these variables only.
Note that no locking is needed since global_tx_head is updated in the xmit
flow and global_tx_tail is updated in the NAPI flow only. A previous
attempt tried to use one variable to count the outstanding sent packets,
but it did not work since xmit and NAPI flows can run at the same time and
the counter will be updated wrongly. Thus, we use the same simple cyclic
head and tail scheme that we have today for the UD tx_ring.
Fixes: 2c104ea683 ("IB/ipoib: Get rid of the tx_outstanding variable in all modes")
Link: https://lore.kernel.org/r/20200527134705.480068-1-leon@kernel.org
Signed-off-by: Valentine Fatiev <valentinef@mellanox.com>
Signed-off-by: Alaa Hleihel <alaa@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Acked-by: Doug Ledford <dledford@redhat.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
The most common way to set ECE option will be during modify QP command in
INIT2RTR, RTR2RTS and RTS2RTS stages, so update mlx5 to support it.
The new bit in the comp_mask is needed to mark that kernel supports ECE
and can receive data instead of "reserved" field in the struct
mlx5_ib_modify_qp.
Link: https://lore.kernel.org/r/20200526115440.205922-8-leon@kernel.org
Reviewed-by: Mark Zhang <markz@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
IBTA declares "vendor option not supported" reject reason in REJ messages
if passive side doesn't want to accept proposed ECE options.
Due to the fact that ECE is managed by userspace, there is a need to let
users to provide such rejected reason.
Link: https://lore.kernel.org/r/20200526103304.196371-7-leon@kernel.org
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
The rdma_accept() is called by both passive and active sides of CMID
connection to mark readiness to start data transfer. For passive side,
this is called explicitly, for active side, it is called implicitly while
receiving REP message.
Provide ECE data to rdma_accept function needed for passive side to send
that REP message.
Link: https://lore.kernel.org/r/20200526103304.196371-6-leon@kernel.org
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Passive side of CMID connection receives ECE request through REQ message
and needs to respond with relevant REP message which will be forwarded to
active side.
The UCMA events interface is responsible for such communication with the
user space (librdmacm). Extend it to provide ECE wire data.
Link: https://lore.kernel.org/r/20200526103304.196371-4-leon@kernel.org
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
From the mlx5-next branch at
git://git.kernel.org/pub/scm/linux/kernel/git/mellanox/linux
Required for dependencies in following patches
* branch 'mellanox/mlx5-next':
net/mlx5: Add ability to read and write ECE options
net/mlx5: Add support for RDMA TX FT headers modifying
net/mlx5: Move iseg access helper routines close to mlx5_core driver
net/mlx5: Cleanup mlx5_ifc_fte_match_set_misc2_bits
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
